I build AI-native systems and backend software where outputs, data and system behaviour remain inspectable.
The pinned repositories are the work. This profile is the context behind it.
I care about systems that remain understandable after they ship: where a fact came from, who owns the state, what happens when a dependency fails and how another engineer can reproduce or challenge the result.
| Engineering concern | What I look for |
|---|---|
| AI behaviour | Grounded context, citations where evidence matters and clear output boundaries |
| Backend systems | Typed contracts, explicit ownership, durable state and defensible data decisions |
| Reliability | Failure paths, controlled errors, tests that exercise more than the happy path |
| Trust | Provenance, audit history, deterministic checks and honest capability limits |
| Delivery | Readable repositories, meaningful pull requests, CI, review gates and documentation |
- Repository intelligence: building PARTHA, which converts a codebase into structured, source-backed context for engineers and AI tools.
- Open-source quality: holding merge access to OpenShield's
devbranch and owning its tests and documentation as a CODEOWNER. - AI products: building systems that keep retrieval, model outputs and state boundaries checkable rather than opaque.
Python · TypeScript · JavaScript · SQL · C++
FastAPI · Django · Django Ninja · Flask · REST APIs · Pydantic · PostgreSQL · Redis · MongoDB · SQLite · SQLAlchemy · Alembic
LLM integrations · Workers AI · RAG · Embeddings · Retrieval · Citations · Output validation · Model evaluation · MCP
Cloudflare Workers · Durable Objects · Microsoft Azure · Docker · GitHub Actions · CI/CD · Application Insights
pytest · Django tests · Vitest · Playwright · CodeQL · Dependency audits · Failure-path testing · RBAC
I contribute to OpenShield, an open-source Azure cloud-security posture management project.
My work includes scanner correctness, compliance mappings, database reliability, CI validation, tests, documentation and the OpenShield Learn portal. The project holds an OpenSSF Best Practices passing badge and is progressing through OWASP governance. I will become a maintainer under OWASP when that migration completes.
I use Codex and Claude Code to accelerate exploration and implementation. The architecture, tests, reviews and code that leave a repository remain my responsibility.